Association of PIK3CA Mutation Status before and after Neoadjuvant Chemotherapy with Response to Chemotherapy in Women with Breast Cancer

Approximately 30% of breast cancer patients carry PIK3CA (phosphatidylinositol-4, 5-bisphosphate 3-kinase, catalytic subunit alpha) mutations (1–6). More than 90% of the PIK3CA mutations in breast cancers occur in three hotspots of the PIK3CA gene: E542 and E545 in the helical domain (exon 9) and H1047 in the kinase domain (exon 20; refs. 2, 7–11). All three hotspot mutations are oncogenic and constitutively activate the PI3K pathway (3, 12–20). The remaining mutations are distributed over the entire PIK3CA-coding sequence.

The role of PIK3CA mutations in breast cancer survival is controversial (21–31). Many studies suggested that PIK3CA mutations are associated with favorable survival (21–26), others found that PIK3CA mutations are associated with worse outcomes (27, 28), and a number of studies found no association between PIK3CA mutations and breast cancer survival (29–31).

Recent in vitro studies suggested that tumor cells with an activated PI3K pathway are less sensitive to chemotherapy agents (32, 33). However, few studies have investigated the association between PIK3CA mutations and the response to neoadjuvant chemotherapy in breast cancer (19, 34, 35). Therefore, whether breast cancer patients with PIK3CA mutations are sensitive to neoadjuvant chemotherapy is not fully determined.

In the current study, the primary objective was to investigate the association between PIK3CA mutations and the response to neoadjuvant chemotherapy in operable primary breast cancer patients who received neoadjuvant chemotherapy (n = 729), and to explore the association between the PIK3CA mutations and survival in this cohort. The second objective was to identify the PIK3CA mutations in breast cancer tissues before and after neoadjuvant chemotherapy in a subgroup of patients (n = 102) in this cohort and to investigate whether changes in PIK3CA mutations during treatment influenced the clinical outcome of this subgroup.

Fresh breast cancer tissues obtained from a 14-gauge core-needle biopsy prior to neoadjuvant chemotherapy were available for 774 operable primary breast cancer patients (stages I–III) who received neoadjuvant chemotherapy at the Breast Center, Peking University Cancer Hospital (Beijing, China) from May 2004 to May 2011. PIK3CA mutations were successfully identified in fresh cancer tissues that were taken before neoadjuvant chemotherapy in 729 of 774 patients (94.2%). Of these 729 patients, breast cancer tissues were also available for 102 patients after neoadjuvant chemotherapy, and PIK3CA mutations were successfully identified in these 102 patients.

The patients' ages at diagnosis ranged from 25 to 73 years, with a median of 49 years. The tumors were graded according to the modified Bloom–Richardson system. Tumor stage was classified according to the tumor–node–metastasis (TNM) classification of the Union Internationale Contre le Cancer. Tumor size was defined as the maximum tumor diameter measured by ultrasound at the time of diagnosis.

Follow-up data were available for all 729 patients, the median length of follow-up was 66 months (range, 5–113 months). During the follow-up period, 143 patients experienced a local or distant recurrence or died of the disease. The study was approved by the Research and Ethical Committee of Peking University Cancer Hospital.

Total RNA was extracted from breast tumor tissues obtained from a core-needle biopsy prior to neoadjuvant chemotherapy or taken following surgery using TRizol reagent (Life Technologies Inc.) according to the manufacturer's instructions. In brief, 500 ng of RNA was transcribed to cDNA by a reverse transcriptase in a total of 20 μL reverse transcription reaction solution containing 4.0 μL 5× first strand buffer, 10 mmol/L DTT, 20 U RNase inhibitor, 1 mmol/L dNTP, 400 ng random primer and 200 U superscript II reverse transcriptase (Invitrogen). The resulting cDNA was subjected to PCR amplification. Mutational analysis of PIK3CA was performed using a set of 8 primer pairs (Supplementary Table S1) that covered the entire coding region of the PIK3CA gene. All fragments were sequenced using the BigDye Terminator Cycle Sequencing Kit and ABI3730 automated sequencer (Applied Biosystems). Each mutation was confirmed in duplicate.

Estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor-2 (HER2) statuses were determined using the breast cancer tissues obtained from the core-needle biopsy using an immunohistochemical (IHC) assay, as described previously (36). ER or PR was considered positive when ≥1% of tumor cells showed positive nuclear staining. HER2 positivity was defined as a score of 3+ (IHC) or by HER2 gene amplification (fluorescence in situ hybridization). In this study, the breast cancer subgroups were classified according to ER, PR, and HER2 statuses: luminal (ER⁺ and/or PR⁺, HER2⁻), HER2⁺, and triple-negative (TN, ER⁻/PR⁻/HER2⁻; ref. 30).

All the 729 patients received neoadjuvant chemotherapy, and 93% of the patients received 4 to 8 cycles. Treatments were categorized into four subgroups as follows:

1. A total of 265 patients received anthracycline based regimens. Of these, 184 patients received a 5-fluorouracil, pirarubicin, and cyclophosphamide (CTF) regimen, the details of which are described previously (36); 56 patients received the FEC regimen, epirubicin 90 to 100 mg/m² on day 1, every 3 weeks, the doses of 5-fluorouracil and cyclophosphamide were identical to the CTF regimen; 23 patients received the CAF regimen; 1 patient received pirarubicin plus cyclophosphamide regimen; and 1 patient received the doxorubicin and cyclophosphamide (AC) regimen.

2. A total of 317 patients received anthracycline–taxane-based regimens. Of these, 131 patients received 2 cycles of anthracycline regimens followed by 4 cycles of paclitaxel alone (80 mg/m² i.v. once a week for 12 weeks) or docetaxel plus cyclophosphamide (docetaxel 75 mg/m² i.v. on day 1, cyclophosphamide 600 mg/m² i.v. on day 1, every 3 weeks). One hundred and eighty-six patients received 2 cycles of anthracycline regimens followed by paclitaxel plus carboplatin (paclitaxel 175 mg/m² i.v. on day 1, or paclitaxel 60 mg/m² i.v. on days 1, 8, and 15, carboplatin AUC6, i.v. on day 1, every 3 weeks).

3. Taxane-based regimens: 100 patients received 4 cycles of paclitaxel alone (paclitaxel 60 mg/m² i.v. on days 1, 8, and 15, every 3 weeks).

4. Other regimens: the remaining 47 patients in this cohort of 729 patients received other regimens.

A total of 222 patients had HER2⁺ tumors. Of these, 41 women were treated with neoadjuvant trastuzumab in combination with one of the above-described regimens. pCR was defined as no invasive breast cancer cells in the breast after completion of neoadjuvant chemotherapy (37).

Approximately 59.2% of the patients received adjuvant chemotherapy after surgery. Of these, 55.3% of patients with PIK3CA mutations and 60.6% of patients with wild-type received adjuvant chemotherapy after operation, respectively. In addition, patients with axillary positive lymph nodes (≥3 nodes) or who had breast-conserving surgery received radiotherapy, and patients with ER- and/or PR-positive disease received endocrine therapy.

The associations between PIK3CA mutations and clinicopathologic characteristics or pCR rate were performed using the Pearson χ² test or the Fisher exact test. For the survival analyses, disease-free survival (DFS) was defined as the time from the date of diagnosis to first recurrence (local or distant), the contralateral breast cancer, or death from breast cancer without a recorded relapse. Distant disease-free survival (DDFS) was defined as the time from the date of diagnosis to either the first distant recurrence or death for which breast cancer was the primary or underlying cause. Survival was estimated using the Kaplan–Meier product-limit method, and differences were tested for statistical significance using the log-rank test. Cox proportional hazards regression models were used to test the prognostic role of the PIK3CA mutation status [HR and 95% confidence intervals (CI)]. Two-sided P values less than 0.05 were considered to be statistically significant. All analyses were performed using the SPSS Statistics 20.0 software.

Of these patients, 28.3% (206/729) had PIK3CA mutations. The frequencies of PIK3CA mutations in the luminal, HER2⁺, and TN subgroups were 30.6%, 27.5%, 22.7%, respectively (P = 0.22; Supplementary Table S2). PIK3CA mutant than wild-type were more likely to be ER- and PR-positive (ER: P = 0.034; PR: P = 0.007, respectively), and no significant association was found between PIK3CA mutations and other clinicopathologic characteristics (Table 1).

Six patients contained two PIK3CA mutations; therefore, 212 PIK3CA mutations were found in the 206 patients. Of these 212 mutations, 87.3% (185/212) clustered in the three hotspots (E542/E545 and H1047), and 12.7% (27/212) clustered in non-hotspots.

In total, 19.5% of patients (142/729) achieved a pCR after neoadjuvant chemotherapy. Patients with PIK3CA mutations had a lower pCR rate than those with wild-type (pCR rate, 14.6% vs 21.4%, P = 0.035; Table 2). In multivariate analysis, PIK3CA mutation status tended to be an unfavorable factor for pCR (OR, 0.68; 95% CI, 0.42–1.11; P = 0.13; Supplementary Table S3). Patients with PIK3CA mutations located in the three hotspots (E542/E545 and H1047) had a lower pCR rate than patients with wild-type (13.5% vs 21.4%, P = 0.019; Supplementary Table S4). No significant difference in pCR rate between patients with non-hotspot mutations and patients with wild-type was observed (23.8% vs. 21.4%, P = 0.79; Supplementary Table S4). The pCR rate for patients with E542/E545 mutations was similar to those with H1047 mutations (12.5% vs 14.0%, P = 0.68; Supplementary Table S5). However, patients with H1047 mutation showed a trend for better survival than those with E542/E545 mutation (DFS: unadjusted HR, 0.71; 95% CI, 0.39–1.29, P = 0.25; DDFS: unadjusted HR, 0.60; 95% CI, 0.31–1.16, P = 0.13; Supplementary Fig. S1). When the pCR includes both breast and axillary lymph nodes, PIK3CA mutations were associated with lower pCR rate in the entire study cohort (mutant vs. wild-type, 13.6% vs. 18.2%, P = 0.14), although the difference did not reach a significance.

Of the patients treated with anthracycline-based regimens (n = 265), patients with PIK3CA mutations had a lower pCR rate than those with wild-type (14.1% vs. 19.4%, P = 0.33; Table 3); among patients treated with anthracyclin–taxane-based regimens (n = 317), patients with PIK3CA mutations had a significantly lower pCR rate than those with wild-type (11.7% vs. 24.7%, P = 0.009; Table 3); among patients treated with taxane-based regimen (n = 100), the pCR rate was similar in patients with or without PIK3CA mutations (16.1% vs. 18.8%, P = 0.74; Table 3).

In the luminal subgroup (n = 376), patients with PIK3CA mutations had a lower pCR rate than those with wild-type (6.1% vs. 13.0%, P = 0.047; Supplementary Table S6). In the TN subgroup (n = 128), patients with PIK3CA mutations had a lower pCR rate than those without mutations (27.6% vs 37.4%, P = 0.33), although the difference did not reach a significance (Supplementary Table S6). The pCR rate was identical between PIK3CA mutant and wild-type among the HER2⁺ patients who did not receive trastuzumab treatment (n = 181; 20.0% vs. 19.9%, P = 0.98); among the HER2⁺ patients who received neoadjuvant trastuzumab (n = 41), patients with PIK3CA mutations had a lower pCR rate than those with wild-type (37.5% vs. 56.0%), but this difference was not significant (P = 0.25; Supplementary Table S6).

The PIK3CA mutation status before and after neoadjuvant chemotherapy in 102 patients in this cohort was assessed. Of these, 24 patients (23.5%) had PIK3CA mutations before neoadjuvant chemotherapy (detailed information on these patients is presented in Table 4). After neoadjuvant chemotherapy, of these 24 patients, 15 patients retained their initial PIK3CA mutations (referred to as the mt-mt group), and these 15 patients mainly exhibited luminal subtype (Supplementary Table S7); 9 patients lost their initial mutations (referred to as the mt-wt group; Table 4). The PIK3CA status remained stable in the 78 patients with wild-type after neoadjuvant chemotherapy (referred to as the wt-wt group). Patients who retained their initial mutations had a lower pCR rate than those who lost their initial mutations (0.0% vs. 33.0%, P = 0.042; Table 5). Patients who retained their initial mutations after neoadjuvant chemotherapy had a borderline worse survival than those who lost their mutations or those with wild-type (DFS: unadjusted HR, 2.32; 95% CI, 0.97–5.57; P = 0.052; and DDFS: unadjusted HR, 2.34; 95% CI, 0.98–5.62; P = 0.050; Fig. 1). In multivariate analysis, retained the initial mutation remained as a nonsignificantly unfavorable factor (DFS: adjusted HR, 2.27; 95% CI, 0.83–6.25; P = 0.11; and DDFS: adjusted HR, 1.32; 95% CI, 0.92–1.90; P = 0.14; Supplementary Table S8).

The 5-year DFS and 5-year DDFS rates in the entire study population (n = 729) were 81.3% and 84.4%, respectively. The 5-year DFS in PIK3CA mutant and wild-type were 79.4% and 82.0%, respectively, and the 5-year DDFS in PIK3CA mutant and wild-type were 82.8% and 85.0%, respectively. In the univariate analysis, there were no significant differences in DFS and DDFS between the PIK3CA mutant and wild-type in the entire study population (DFS: unadjusted HR, 1.25; 95% CI, 0.88–1.77; P = 0.21; and DDFS: unadjusted HR, 1.18; 95% CI, 0.81–1.74; P = 0.39; Supplementary Fig. S2).

The association between the PIK3CA mutations and survival in different subgroups was also analyzed. Patients with PIK3CA mutations had a slightly worse survival than those with wild-type in the luminal subgroup (DFS: unadjusted HR, 1.30; 95% CI, 0.76–2.23; P = 0.33; and DDFS: unadjusted HR, 1.55; 95% CI, 0.89–2.69; P = 0.12; Supplementary Fig. S3).

Among the HER2⁺ patients (HER2⁺ subgroup) who did not receive trastuzumab treatment, PIK3CA mutant had a slightly better survival than wild-type (DFS: unadjusted HR, 0.82; 95% CI, 0.42–1.60; P = 0.55; and DDFS: unadjusted HR, 0.51; 95% CI, 0.21–1.22; P = 0.13; Supplementary Fig. S4). Among the HER2⁺ patients who received trastuzumab treatment, patients with PIK3CA mutations had a tendency toward worse survival than those with wild-type (DFS: unadjusted HR, 3.34; 95% CI, 0.61–18.27; P = 0.14; DDFS: unadjusted HR, 4.73; 95% CI, 0.49–45.46; P = 0.14; Supplementary Fig. S5).

In the TN subgroup, PIK3CA mutant had a significantly worse DFS and non-significantly worse DDFS than wild-type (DFS: unadjusted HR, 2.11; 95% CI, 1.00–4.43; P = 0.044; and DDFS: unadjusted HR, 1.69; 95% CI, 0.73–3.91; P = 0.22; Supplementary Fig. S6).

To our knowledge, the current study is one of the largest studies to investigate the association between PIK3CA mutations and response to neoadjuvant chemotherapy in breast cancer to date. We found that patients with PIK3CA mutations had a significantly lower pCR rate than those with wild-type. These findings suggested that PIK3CA mutant are resistant to neoadjuvant chemotherapy.

We explored the PIK3CA mutation status in 102 patients before and after neoadjuvant chemotherapy. Patients who retained their initial PIK3CA mutations showed a worse survival than those who lost their initial mutations or those with wild-type. These findings suggested that patients who retained their initial mutations after neoadjuvant chemotherapy harbored an aggressive phenotype and were less sensitive to chemotherapy; therefore, those patients might be suitable for PIK3CA-targeted therapy, as one recent study indicated that breast or gynecologic cancer patients with PIK3CA mutations have a higher response to PI3K/AKT/mTOR inhibitors than those with wild-type (11). On the other hand, patients who lost their initial mutations after neoadjuvant chemotherapy had a favorable outcome and benefited from neoadjuvant chemotherapy.

The three hotspot mutations (E542/E545 and H1047) accounted for the majority of mutations of the PIK3CA gene in the current study. The mechanisms that activate the PI3K/AKT pathway may be different between the E542/E545 mutations and the H1047 mutation (13–17,23, 24). Janku and colleagues recently reported that patients with H1047 mutation are more sensitive to a PI3K/AKT/mTOR inhibitor than those with other mutations or patients with wild-type, and patients with H1047 mutation show a trend for a longer progression-free survival than those with other mutations (38, 39). In our present study, although the E542/E545 and H1047 mutations exhibited a similar efficacy in response to neoadjuvant chemotherapy, patients with H1047 mutation had a nonsignificantly better survival than those with E542/E545 mutation.

Emerging evidence suggested that PIK3CA mutations have different roles in different subgroups when stratified by ER, PR, and HER2 status (25, 29, 40, 41). We found that the frequency of PIK3CA mutations was higher for luminal subgroup than for the HER2⁺ and TN subgroups, and this observation is consistent with those of previous studies (24, 26). PIK3CA mutations were significantly associated with a lower pCR rate in the luminal subgroup, whereas PIK3CA mutations were associated with a lower pCR rate in the TN subgroup, although the difference was not significant. PIK3CA mutations were not associated with pathologic response among the HER2⁺ patients treated with neoadjuvant chemotherapy without trastuzumab. Our results suggested that the association between mutated PIK3CA and a reduced pCR rate is dependent on the molecular types, and the association is more pronounced in the luminal subgroup.

In contrast, the HER2⁺ patients treated with neoadjuvant chemotherapy in combination with trastuzumab, we found that patients with PIK3CA mutations had a lower pCR rate than those with wild-type. Interestingly, two recent studies have demonstrated that HER2⁺ breast cancer patients with PIK3CA mutations are less likely to achieve a pCR compared with those with wild-type when the patients treated with neoadjuvant chemotherapy plus anti-HER2 therapies in the GeparQuattro, GeparQuinto, and GeparSixto trials (19) and the NeoATTO trial (35). Similarly, PIK3CA mutation is associated with a shorter progression-free survival in HER2⁺ metastatic breast cancer treated with HER2-targeted therapies in the CLEOPATRA trial (42). Our present observation is in agreement with these findings. However, unlike the results observed in the neoadjuvant and metastatic settings, there is no association between the PIK3CA mutation and the degree of benefit from transtuzumab in adjuvant setting in the NSABP B-31 trial (43).

Although patients with PIK3CA mutations had a slightly poorer DFS and DDFS than those with wild-type in the entire study population (n = 729), the differences did not reach a significance. However, PIK3CA mutations were associated with worse DFS in the TN subgroup. Due to the relatively small sample size of the TN subgroup (n = 128), further independent studies are warranted to confirm this observation.

There are two limitations to our study. Although the overall cohort was large, the number of patients who had matched tumor tissues before and after neoadjuvant chemotherapy was not large (n = 102); in particular, the PIK3CA mutant in this group were relatively small. The current study was retrospective, and the neoadjuvant chemotherapy regimens were not assigned at randomized. Therefore, caution is required when interpreting our results.

In summary, patients with PIK3CA mutations are less likely to respond to neoadjuvant chemotherapy. Although PIK3CA mutations are not associated with survival in the entire study population, patients who retain their initial PIK3CA mutations after neoadjuvant chemotherapy have a lower pCR rate and an unfavorable survival when compared with those who lose their initial PIK3CA mutations.

No potential conflicts of interest were disclosed.

Conception and design: Y. Xie

Development of methodology: H. Yuan, J. Chen, Y. Liu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Ouyang, T. Wang, Z. Fan, T. Fan, B. Lin, Y. Xie

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Yuan, Y. Liu, Y. Xie

Writing, review, and/or revision of the manuscript: H. Yuan, Y. Xie

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Ouyang, J. Li, T. Wang, Y. Xie

Study supervision: J. Li, Y. Xie

This study was supported by the National Key Technology Research and Development Program of the Ministry of Science and Technology of China (No. 2014BAI09B08), the 973 project 2013CB911004, the 985-III project, and grants from the National Natural Science Foundation of China (Nos. 30973436 and 81071629).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.